The Impact of Urbanization in Indonesia: Analysis of Firm Productivity and Labor Migration

IDE Research Bulletin

March 2019

PDF (604KB)

Brief summary of the project

This project consists of two studies. “Firm Agglomeration and Aggregate Fluctuations” examines the role of individual firms located in agglomeration areas in generating aggregate fluctuations. We used two sets of balanced panel data: (1) the balanced panel from 1996 to 2014, and (2) the balanced panel from 2006 to 2014. The methodology is based on the decomposition method proposed by Giovanni et al. (2014), which enabled us to decompose aggregate fluctuations into (sector-level) macroeconomic shocks and firm-level shocks.

Our main findings are as follows. First, like Giovanni et al.’s (2014) empirical results, firm-level shocks mainly contribute to aggregate fluctuations instead of sector-level macroeconomic shocks. Table 1 shows that firm-specific shocks contribute more than 80% to the aggregate fluctuations. Second, firm-to-firm linkages play an important role in explaining the magnitude of firm-specific shocks. As column (1) of Table 2 shows, the covariance of shocks among firms (LINK) is much greater than the variance of individual shocks (DIRECT). Third, we decomposed the effect of firm-specific shock into two groups: (1) firms located in higher agglomeration areas and (2) firms in lower agglomeration areas. We found that firms in higher agglomeration areas have greater firm-specific fluctuations (Table 2). Fourth, a comparison of DIRECT and LINK effects, as seen in Table 2, revealed that a relatively large difference between the higher and lower agglomeration groups lies in the LINK effect. Therefore, aggregate fluctuations in the Indonesian manufacturing sector are mainly caused by fluctuations of firms located in higher agglomeration areas. In particular, the co-movement among those firms is a key driver of aggregate fluctuations.

In “The Effects of Internal Migration on the Labor Market in Indonesia,” we analyze the effects of migration on the labor market outcomes of native residents. We treated natural geological disasters such as earthquakes, tsunamis, and volcanic eruptions from 2005 to 2008 as quasi-experimental events that lead to a pushing out of migrants from those districts, to identify the causal effects of migration on natives who lived in their communities at least from 2000 to 2014. In order to analyze the effects on labor outcome, we used a panel dataset of individuals from the Indonesian Family Life Survey after merging the data with our original urban area dataset by community (Figure 1).

Tables 3 and 4 show the main results of our benchmark analysis; a 10km radius around communities was regarded their labor market. We found that the predicted share of migration on employment is not statistically significant after we control for community-level characteristics. On the other hand, as shown in Table 4, we found that the effects of inflow of migrants on income of individuals are statistically significant if natives live in urban areas. According to columns (8), a one-percentage-point change of the migration share increases the income of natives living in urban areas by around 7%. We also checked the robustness of our estimation by 1) changing the labor market size from a 5km radius to 20km, (2) analyzing only labor outcomes of males, and (3) using different weights for IV. We found positive and statistically significant effects of migration on the income of natives in urban areas in the analysis of robustness check.